Political accountability and sequential policymaking

We develop a model of political accountability with sequential policy making. Consistent with empirical observations, equilibrium behavior by the policymaker overemphasizes the late stages of the policy-making process. The reason is that the overseer faces a political time inconsistency problem—she is tempted to revise her retention rule in the middle of the policy-making process. If the overseer knows the technology by which policies translate into outcomes, then she can eliminate these distortions using taskspecific budget caps. However, if the overseer is uncertain about this technology, such budget caps introduce ex post inefficiency. If the uncertainty is sufficiently large and consequential, the overseer prefers an institutional environment in which policymaker actions are non-transparent and the budget is fungible. Such an environment allows the overseer to exploit the policymaker’s expertise about the technology, but at the cost of weaker overall incentives. Hence, the model highlights a novel mechanism by which transparency may not always be optimal in political environments. ∗This paper replaces and subsumes our earlier paper, “Does Clarifying Responsibility Always Improve Policy?” We have received helpful comments from Scott Ashworth, Cathy Hafer, and Richard van Weelden. This research was supported by NSF grants SES-0819152 (Bueno de Mesquita) and SES-0819545 (Landa). †Professor, Harris School of Public Policy Studies, University of Chicago, e-mail: [email protected] ‡Associate Professor, Wilf Family Department of Politics, NYU, e-mail: [email protected] How does the sequential nature of the policy-making process impact the efficacy of political accountability and optimal political institutions? Political economy models typically assume policymakers have only a single opportunity to set policy prior to moments of accountability.1 However, in most important policy domains, a policymaker take multiple, sequential actions between decisions by the overseer regarding whether to reward or punish the policymaker. Moreover, empirical scholarship suggests the presence of important inter-temporal distortions in the behavior of policymakers. Most prominent among these is the well known pattern of behavior whereby overseers primarily focus their attention on the later stages of the policy-making process and policymakers, consequently, allocate disproportionately more effort or resources to these later stages relative to the earlier stages (Popkin et al., 1976; Shepsle and Weingast, 1981; Weingast, Shepsle and Johnsen, 1981; Figlio, 2000; Rothenberg and Sanders, 2000; Albouy, 2011). Such dynamic effects cannot be captured in the context of a single-action model, and perhaps not surprisingly, they have resisted political economy explanations relying on rational agency.2 We develop a model of political agency with sequential actions. Our model yields two key kinds of results that underscore the importance of explicitly modeling sequential policy making. First, our model gives rise to equilibrium behavior by policymakers that overemphasizes the late stages of the policy-making process. The intuition behind this result is a kind of political time inconsistency problem. The overseer, at the beginning of the game, would like to commit to a retention rule that induces an optimal division of resources across tasks. However, following the early stages of the policy-making process, the overseer is tempted to revise her retention rule in order to optimize incentives going forward. The policymaker anticipates this incentive to revise and consequently underemphasizes the early stages of the policy-making process. Second, we provide a novel argument for the potential benefits of eliminating transparency in political settings. This argument starts with the observation that, if the overseer knows the technology by which policy translates into outcomes, then she can solve the time inconsistency problem (and the distortions it induces) by establishing perfectly tailored Of course, multitask problems (Holmström and Milgrom, 1991) have been analyzed in many settings, including political agency settings (Lohmann, 1998; Besley and Coate, 2003; Ashworth, 2005; Ashworth and Bueno de Mesquita, 2006; Gehlbach, 2006; Bueno de Mesquita, 2007; Bueno de Mesquita and Stephenson, 2007; Hatfield and Padró i Miquel, 2007; Daley and Snowberg, 2011; Ashworth and Bueno de Mesquita, 2014; Le Bihan, 2014). None of these models consider the issue of sequential actions prior to a retention decision, which is our focus. The only other papers we are aware of that yield such behavior in equilibrium are Sarafidis (2007) and Muthoo and Shepsle (2010). In these models, the overemphasis of late stages of the policy-making process emerges due to the assumption that the overseers have a “recency bias”, whereas in our model it is a result of purely strategic factors with fully rational overseers.